Seasonal analysis and prediction of CARFI index based on CNN

The China Automobile Roll-on/Roll-off Freight Index (CARFI) is a pivotal metric for shipping enterprises navigating China's inland roll-on/roll-off transportation market. This study introduces a Convolutional Neural Network (CNN) model designed to forecast weekly indices by analyzing historical time series data and isolating seasonal components. Seasonal analysis and subsequent predictions are performed, and the model's efficacy is compared with the traditional ARIMA forecasting model. The results demonstrate that the CNN model not only facilitates seasonal analysis but also delivers more precise predictions than the ARIMA model. Employing the CNN model for forecasting the CARFI index enables enterprises to proactively formulate strategies in response to market fluctuations, thereby promoting sustainable market development.